The sense of agency in human–AI interactions
Roberto Legaspi,Wenzhen Xu,Tatsuya Konishi,Shinya Wada,Nao Kobayashi,Yasushi Naruse,Yuichi Ishikawa
DOI: https://doi.org/10.1016/j.knosys.2023.111298
IF: 8.139
2024-02-01
Knowledge-Based Systems
Abstract:Sense of agency (SoA) is the perceived control over one’s actions and their consequences, and through this one feels responsible for the consequent outcomes in the world. We analyze the far-reaching implications of a two-pronged knowledge on SoA and its impact on human–AI interactions. We argue that although there are interesting research efforts for an AI to inherently possess SoA, they are still sparse, constrained in scope and present unclear immediate benefit to the design of AI-enabled systems. We also argue that the knowledge on how human SoA is affected by an AI that is perceived to possess a sense of control presents more immediate benefit to AI, in particular, to eliciting positive human attitudes toward AI. Third, and lastly, we argue that research efforts for an AI to adapt to the dynamic changes of human SoA are practically non-existent primarily due to the difficulty of modeling, inferring and adaptively responding to human SoA in complex natural settings. We proceed by first delving deep into the influential and recent theoretical underpinnings of SoA, and discuss its conceptual reach in different disciplines and how it is applied in real-world research. We organize a substantial part of our paper to put forward and elucidate our three argumentative points while supported by evidence in the literature.
computer science, artificial intelligence